path: root/src/docs/wtperf.dox

/*! @page wtperf Simulating workloads with wtperf

The WiredTiger distribution includes a tool that can be used to simulate
workloads in WiredTiger, in the directory \c bench/wtperf.

The \c wtperf utility generally has two phases, the populate phase which
creates a database and then populates an object in that database, and a
workload phase, that does some set of operations on the object.

For example, the following configuration uses a single thread to
populate a file object with 500,000 records in a 500MB cache.  The
workload phase consists of 8 threads running for two minutes, all
reading from the file.

@code
conn_config="cache_size=500MB"
table_config="type=file"
icount=500000
run_time=120
populate_threads=1
threads=((count=8,reads=1))
@endcode

The \c conn_config configuration supports setting any WiredTiger
connection configuration value.  This is commonly used to configure
statistics with regular reports, to obtain more information from the
run:

@code
conn_config="cache_size=20G,statistics=(fast,clear),statistics_log=(wait=600)"
report_interval=5
@endcode

Note that quoting must be used when passing values to Wiredtiger
configuration as opposed to configuring the \c wtperf utility.

The \c table_config configuration supports setting any WiredTiger object
creation configuration value, for example, the above test can be
converted to using an LSM store instead of a B+tree store, with
additional LSM configuration by changing \c conn_config to:

@code
table_config="lsm=(chunk_size=5MB),type=lsm,os_cache_dirty_max=16MB"
@endcode

More complex workloads can be configured by creating more threads doing
inserts and updates as well as reads.  For example, to configure two
inserting threads, two reading threads and two updating threads:

@code
threads=((count=2,inserts=1),(count=2,reads=1),(count=2,updates=1))
@endcode

Some of the standard \c wtperf configurations we run can be found in the
\c bench/wtperf/runners/ directory.

@section monitor Monitoring performance

In addition to testing a particular configuration, \c wtperf can monitor
performance and operation latency times.  Monitoring can be enabled using
the \c sample_interval configuration setting.  For example to record
every 10 seconds, add the following to the \c wtperf configuration file:

@code
sample_interval=10
@endcode

Enabling monitoring causes \c wtperf to create a file \c monitor in
the database home directory, or another directory as specified in
the \c -m option to \c wtperf.

A visualization tool to see the performance over the course of the
\c wtperf run can be found in \c tools/wtperf_stats.py.

The following example shows how to run the \c fruit-short.wtperf
stock configuration to a subdirectory of \c build_posix called
\c WT_TEST with monitoring enabled and then generate a graph.
This example assumes you have all the necessary tools
already installed to build WiredTiger, Python and python-nvd3.

@code
mkdir WT_TEST && ./bench/wtperf/wtperf -O ../bench/wtperf/runners/fruit-short.wtperf -o "sample_interval=10"
cd WT_TEST; python ../../tools/wtperf_stats.py monitor
@endcode

Once the python tool completes, you will find a file \c wtperf_stats.html
that will look something like this:

\htmlinclude wtperf-sample.html

@section config Configuration options

The following is a list of the currently available \c wtperf
configuration options:

\if START_AUTO_GENERATED_WTPERF_CONFIGURATION
DO NOT EDIT: THIS PART OF THE FILE IS GENERATED BY dist/s_docs.
\endif

@par async_threads (unsigned int, default=0)
number of async worker threads
@par checkpoint_interval (unsigned int, default=120)
checkpoint every interval seconds during the workload phase.
@par checkpoint_stress_rate (unsigned int, default=0)
checkpoint every rate operations during the populate phase in the
populate thread(s), 0 to disable
@par checkpoint_threads (unsigned int, default=0)
number of checkpoint threads
@par conn_config (string, default=create)
connection configuration string
@par compact (boolean, default=false)
post-populate compact for LSM merging activity
@par compression (string, default=none)
compression extension.  Allowed configuration values are: 'none',
'bzip', 'snappy', 'zlib'
@par create (boolean, default=true)
do population phase; false to use existing database
@par database_count (unsigned int, default=1)
number of WiredTiger databases to use. Each database will execute the
workload using a separate home directory and complete set of worker
threads
@par icount (unsigned int, default=5000)
number of records to initially populate. If multiple tables are
configured, each table has this many items inserted.
@par insert_rmw (boolean, default=false)
execute a read prior to each insert in workload phase
@par key_sz (unsigned int, default=20)
key size
@par min_throughput (unsigned int, default=0)
abort if any throughput measured is less than this amount.  Requires
sample_interval to be configured
@par max_latency (unsigned int, default=0)
abort if any latency measured exceeds this number of
milliseconds.Requires sample_interval to be configured
@par pareto (boolean, default=false)
use pareto 80/20 distribution for random numbers
@par populate_ops_per_txn (unsigned int, default=0)
number of operations to group into each transaction in the populate
phase, zero for auto-commit
@par populate_threads (unsigned int, default=1)
number of populate threads, 1 for bulk load
@par random_range (unsigned int, default=0)
if non zero choose a value from within this range as the key for
insert operations
@par random_value (boolean, default=false)
generate random content for the value
@par report_interval (unsigned int, default=2)
output throughput information every interval seconds, 0 to disable
@par run_ops (unsigned int, default=0)
total read, insert and update workload operations
@par run_time (unsigned int, default=0)
total workload seconds
@par sample_interval (unsigned int, default=0)
performance logging every interval seconds, 0 to disable
@par sample_rate (unsigned int, default=50)
how often the latency of operations is measured. One for every
operation,two for every second operation, three for every third
operation etc.
@par sess_config (string, default=)
session configuration string
@par table_config (string, default=key_format=S,value_format=S,type=lsm,exclusive=true,allocation_size=4kb,internal_page_max=64kb,leaf_page_max=4kb,split_pct=100)
table configuration string
@par table_count (unsigned int, default=1)
number of tables to run operations over. Keys are divided evenly over
the tables. Default 1, maximum 99.
@par threads (string, default=)
workload configuration: each 'count' entry is the total number of
threads, and the 'insert', 'read' and 'update' entries are the ratios
of insert, read and update operations done by each worker thread;
multiple workload configurations may be specified; for example, a more
complex threads configuration might be
'threads=((count=2,reads=1)(count=8,reads=1,inserts=2,updates=1))'
which would create 2 threads doing nothing but reads and 8 threads
each doing 50% inserts and 25% reads and updates.  Allowed
configuration values are 'count', 'reads', 'inserts', 'updates'
@par transaction_config (string, default=)
transaction configuration string, relevant when populate_opts_per_txn
is nonzero
@par table_name (string, default=test)
table name
@par value_sz (unsigned int, default=100)
value size
@par verbose (unsigned int, default=1)
verbosity
@par warmup (unsigned int, default=0)
How long to run the workload phase before starting measurements

\if STOP_AUTO_GENERATED_WTPERF_CONFIGURATION
DO NOT EDIT: THIS PART OF THE FILE IS GENERATED BY dist/s_docs.
\endif

*/